On 20 Feb 2013, at 11:47, Simon Slavin wrote:
On 20 Feb 2013, at 5:32pm, Seebs <se...@seebs.net> wrote:
First: The SQL is completely trivial.
Second: I am not having performance problems with sqlite, I am having
performance problems with :memory:. Performance on files is lovely.
This normally means that your computer has very little actual memory
to devote to 'memory' uses. It is using a lot of swap space (or
paging, or virtual memory, whatever you want to call it). So when you
think you are doing calls which reference ':memory:' it actually has
to fetch and write those parts of memory to disk.
I am pretty sure that's not it.
I know I'm a newbie on the list, but please imagine for the sake of
argument that I am a basically competent programmer with enough sysadmin
background to be aware of the obvious problems.
In this case, I tested this pretty carefully over a period of about six
hours of testing across multiple machines. I was watching CPU load,
memory load, and all that stuff. And this is a machine with >4GB of
*free* memory -- that's over and above even the couple GB of disk cache
being used.
Process memory size is not appreciably different between sqlite 3.6 and
3.7, or between page size of 1024 or 8192. Runtime is massively
different. I am pretty sure this is an actual computation-time issue,
and my intuition is that it's quite possible there's at least some other
performance issues lurking, because it appears that :memory: *used to
be* dramatically faster than disk, but something changed in the last
couple of years.
-s
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
